1. Field of the Invention
The present invention relates to a monocrystal pulling-up system which pulls up and grows a monocrystal of silicon etc. by the Czochralski technique (hereinafter referred to as the "CZ technique"), in particular relates to a monocrystal pulling-up system which can pull a large diameter and high quality monocrystal while controlling the oxygen concentration (density) by providing a carrier gas flow controller.
The present invention also relates to a method of exhausting silicon oxide(SiO) produced from the melt silicon in a quartz crucible to the outside of a pulling chamber by controlling the flow of the carrier gas.
2. Description of the Related Art
To produce a silicon monocrystal by the CZ technique, polycrystalline silicon and the necessary dopant(s), for example, P, B, Sb, As, are inserted into a quartz crucible provided rotatably at a bottom of a pulling chamber, the chamber is evacuated to a vacuum, then a heater arranged around the quartz crucible is used to melt the polycrystalline silicon and the dopant(s). A carrier gas is then passed from an upper portion of the chamber to the quartz crucible. At the same time a seed crystal (starting crystal) attached to and supported by a chuck on a pulling shaft is immersed in the melt silicon in the quartz crucible under conditions of a vacuum of 10 to 20 Torr. Then the pulling shaft is pulled up at a predetermined speed while relatively rotating it with respect to the quartz crucible.
U.S. Pat. No. 4,330,362 discloses a pulling system provided with a member (hereinafter called a "heat cap") comprised of a material able to reflect ultraviolet rays above the crucible so as to partially cover the crucible and the melt silicon in the crucible, to thereby block the radiant heat from the melt surface, promote the formation of the monocrystal, and raise the pulling speed and to keep down the concentration of carbon in the monocrystal.
When pulling a silicon monocrystal using the above-mentioned pulling system, however, there are the following disadvantages.
First, the above heat cap can be expected to a certain extent to have the effect of controlling the flow of the carrier gas, for example, argon gas(Ar), being passed to remove the silicon oxide(SiO) produced from the melt and efficiently eliminating the silicon oxide depositing on the inner wall of the top end of the crucible, but silicon oxide ends up depositing and condensing on the top of the heat cap itself. This is a drawback in that it would fall onto the melt silicon surface and thus obstruct the formation of the silicon monocrystal. This is believed to be because the heat cap has as its main object the literal blocking of ultraviolet rays and is not designed with the intention of control of the flow of the argon gas.
Second, the melt surface near the inner peripheral wall of the quartz crucible ends up being covered by the heat cap, so an operator cannot visually inspect the melt surface from a peephole etc. provided in the-pulling chamber. As a result, there is the disadvantage that it is not possible to quickly deal with any heat deformation of the top end of the crucible, recrystallization or deposition of silicon near the inner peripheral wall of the crucible, or any other disadvantages when they occurred.
In addition, there are the following disadvantages when trying to pull a silicon monocrystal for use for the manufacture of a large diameter VLSi device, for example, a diameter of 6 inches, 8 inches, or more.
With a large diameter crystal, the most important thing is the control of the oxygen concentration (density). This is generally classified in the manufacturing process of an LSI device into high oxygen of, for example, 1.55.times.10.sup.18 atoms/cm.sup.3, medium oxygen of, for example, 1.35.times.10.sup.18 atoms/cm.sup.3, and low oxygen of, for example, 1.15.times.10.sup.18 atoms/cm.sup.3. Further, in some cases, extremely high oxygen and extremely low oxygen are demanded and such classifications are selectively used. For example, when using the intrinsic gettering technique utilizing oxygen precipitation, much use is made of monocrystals from a high oxygen to a medium oxygen concentration. On the other hand, when strength and reduction of lattice faults are required, much use is made of monocrystals from a medium oxygen to a low oxygen concentration. Thus, it is necessary to control the variation in the oxygen concentrations in the axial direction of the crystal and in the silicon wafer surface so that the oxygen concentration of the pulled monocrystal becomes in the designated narrow range.
The "oxygen in the crystal" described here means the oxygen dissolving out from the quartz crucible. Almost of the oxygen, for example, 95% of the oxygen, becomes silicon oxide and is exhausted by the carrier gas to the outside of the pulling chamber. Therefore, the following techniques are known for the control of the oxygen concentration in the crystal:
Approach (1): Changing the rotational speed of the crucible so as to control the supply of oxygen from the wall of the quartz crucible. By this technique, if the rotational speed of the crucible is increased, the amount of oxygen of the pulled monocrystal becomes higher. However, if the rotational speed of the crucible is made lower, the temperature fluctuations of the melt become great and crystal faults become easier to occur at a low oxygen concentration. If the rotational speed of the crucible is raised to obtain a high oxygen pulled monocrystal, it is necessary to raise the rotational speed of the pulled crystal along with the same. There is a problem, however, of the resonance point in the case of pulling a pulled crystal by a wire. Further, if the rotational speed of the pulled crystal is made too high, deformation occurs in the monocrystal and there are problems in the control of the diameter of the monocrystal as well. PA1 Approach (2): Control of the pressure of the carrier gas. If the pressure of the carrier gas is increased, the vaporization of the silicon oxide is suppressed, so the amount of oxygen of the pulled monocrystal becomes higher. However, this approach is governed largely by the structure inside the pulling furnace, so not much can be expected in terms of the response of the control of the oxygen concentration. PA1 Approach (3): Spraying carrier gas on the melt silicon surface in the crucible-so as to control the temperature of the melt silicon surface and control the amount of vaporization of the silicon oxide. When the heat cap is used, for example, the carrier gap between the heat cap and the melt silicon surface and the gap between the heat cap and the pulled monocrystal (hereinafter referred to all together as the "bottom gap") are controlled. By this technique, if the bottom gap is made smaller, the temperature of the melt surface falls, so the amount of vaporization of the silicon oxide is held down and as a result the amount of oxygen of the pulled monocrystal becomes higher. This approach is relatively effective to obtain a high oxygen crystal, but if the flow of the carrier gas is increased and the bottom gap is made too small, the carrier gas will strike the melt hard and therefore cause bubbles in the melt. As a result, there are the problems that the crystal growth will no longer be uniform and further the variations in the oxygen concentration in the surface will become greater. PA1 Approach (4): Control of the discharge of the vaporized silicon oxide by the flow of the carrier gas. If the vaporized silicon oxide is efficiently discharged by the carrier gas from the melt silicon surface to outside of the pulling chamber, the vaporization of the silicon oxide is promoted and as a result the oxygen concentration in the melt is lowered and the amount of oxygen in the pulled crystal becomes lower. There is a gas diffusion layer of the vaporized silicon oxide directly above the melt surface. By using the heat cap, the flow rate of the same is increased by the flow of the carrier gas introduced from above the pulling chamber in the narrowed gap between the heat cap and the melt surface and therefore the thickness of the gas diffusion layer is reduced. As a result, the partial pressure of the silicon oxide on the melt surface becomes lower and vaporization of the silicon oxide is promoted, but if a heat cap is used, an opposing phenomenon simultaneously occurs. Further, if the crucible deforms and the melt surface drops, the subsequent oxygen concentration in the crystal will change. PA1 the tubular portion and the constricted diameter portion are integrally formed and the engagement portion is attached detachably to the tubular portion. PA1 a peephole is provided at the outer wall of the pulling chamber on the line connecting the gap between the front end of the cooling means and the front end of the top of the tubular portion of said flow controller and the bottom gap. PA1 defining by a carrier gas branching means a bottom gap of a predetermined size between a circumference of the pulled monocrystal and the surface of the melt and defining a top gap between the crucible and a heat retaining tube provided at the outside of the same, PA1 defining a first flow path through which the carrier gas flows toward the bottom gap between the carrier gas branching means and pulled monocrystal, PA1 defining a second flow path comprised of a flow path of the carrier gas passing through the top gap and a flow path of said carrier gas passing from the first flow path through the bottom gap and then passing between the surface of the silicon melt and a flow controller, PA1 forming the bottom gap and the top gap so that the amount of the carrier gas flowing through the bottom gap becomes greater than the amount of the carrier gas flowing through the top gap, and PA1 exhausting said silicon oxide together with said carrier gas through said second flow path to the outside of the pulling chamber.
In addition to the disadvantages in the approach (4), in the approach (3), if the size of the bottom gap is increased, the effect of the approach (4) becomes stronger and the oxygen concentration rapidly decreases. Therefore, to control the pulled monocrystal to within the target range of oxygen concentration, it is necessary to continuously control the size of the bottom gap precisely.
In this way, when using a heat sink, there is a problem that it is always difficult to set and manage the conditions.
Further, with a silicon monocrystal used for the production of a large diameter VLSi device, it is desirable that the crystal as a whole have the same heat history as much as possible so that the concentration of the oxygen taken into the crystal becomes uniform, even during the subsequent cooling process. Therefore, a heat cap blocking the radiant heat and/or a water cooling tube are provided.
Further, there are phenomena believed to be related to the behavior of clusters of point faults directly on the growth surface having an effect on the pressure resistance of the oxide film of the device. Therefore, Japanese Unexamined Patent Publication (Kokai) No. 3-275586 discloses the production of a crystal with a high oxide film pressure resistance by lowering the pulling speed to 0.5 mm/min or less in a furnace structure with a usual pulling speed of 1.5 mm/min. This is because it is guessed that by lengthening the residence time in the temperature region of over 1300.degree. C. from the crystal growth interface, the faults relating to the pressure resistance of the oxide film diffuse and disappear.
In view of these problems, the present inventor started studies from a completely new viewpoint smashing fixed conceptions about the heat cap disclosed in U.S. Pat. No. 4,330,362 and designed to block (shield) ultraviolet rays, that is, from the viewpoint of a "carrier gas flow controller", and analyzed the state of flow of the carrier gas using computer simulation to find a numerical solution to the Navier-Stokes equation, i.e., a non-linear fluid diffusion equation based on fluid dynamics and thermodynamics.
First, if the state of flow of carrier gas in the case of pulling a monocrystal by a pulling apparatus equipped with the heat cap disclosed in U.S. Pat. No. 4,330,362 (hereinafter referred to as a "closed type heat cap") is considered, the result becomes as shown in FIG. 1 to FIG. 3.
FIG. 1 is a view showing the state of flow of carrier gas in a pulling system equipped with a closed type heat cap, FIG. 2 is a view of the state of flow of carrier gas analyzed by computer simulation of the Navier-Stokes equation, and FIG. 3 is a view of the temperature distribution obtained by analysis by the same computer simulation.
The heat cap 30 shown in FIG. 1 completely partitions the flow path of the carrier gas G into the top (shown by region X) and bottom (shown by region Y) of a pulling chamber, so the carrier gas G introduced from the top of the pulling chamber passes through the narrow bottom gap 33 between the heat cap 30 and the pulled monocrystal 31 and surface of the melt silicon 32 to be increased in speed. By this colliding with the melt surface positioned directly under the bottom gap 33, the temperature of the melt directly under the gap 33 falls, the vaporization of silicon oxide is suppressed, and the melt 32 of the hatched portion 34 shown in FIG. 1 becomes high in oxygen, but on the other hand, the carrier gas forcibly removes the diffusion layer including the silicon oxide from the melt surface, so vaporization of silicon oxide is promoted.
In the other area of the melt 32, however, the degree of contact with the carrier gas G is smaller than with the portion 34 directly under the bottom gap 33, so the melt becomes relatively low in oxygen concentration. Therefore, the distribution of the oxygen concentration of the melt in the crucible becomes nonuniform and there is an adverse effect on the oxygen distribution (ORG) in the silicon wafer surface of the pulled monocrystal 31.
Note that this state is verified by the results of computer simulation shown in FIG. 2 and FIG. 3.
This problem, it may be concluded, derives from the way the carrier gas flows. Since the heat cap partitions the pulling chamber into a top and bottom section, the carrier gas passing through the bottom gap flows in a so-called "squished" manner.
Based on these studies and the results of analyses, the present inventor took note of the "flow-control of the carrier gas" and discovered that if the carrier gas is suitably guided in the pulling chamber, the temperature region directly above the crystal growth interface can be expanded, the control of the oxygen concentration and ORG can be improved, and the condensation and falling of silicon oxide can be prevented and thereby completed the invention disclosed in Japanese Unexamined Patent Publication (Kokai) No. 1-100,086.
A heat-cap disclosed in JPP 1-100,086 comprises a reflector body and projected stops. The reflector consists of a tube and an inclined cylinder provided at a lower portion of the tube with a tip (end) which is reduced in diameter inward. The tube and the inclined cylinder may be formed integrally or together. Projections are provided at the top of the tube and are affixed to the top of a heat retaining member provided around the crucible. The carrier gas is branched by the tube to flow, on one hand, through a gap between the pulling monocrystal and the tube, and, other hand, through a GaP between the tube and the heat retaining member. That is, the heat cap can form a flow path between the tube and the heat holding member. Thus, the heat cap can be called an open-type heat cap.
The heat cap disclosed in JPP 1-100,086 can overcome the disadvantage of U.S. Pat. No. 4,330,362 as a basic idea, but JPP 1-100,086 does not disclose specific conditions. In addition, the heat cap of JPP 1-100,086 requires some improvements.